Tratlus

Full-Stack AI Travel Platform

An AI-assisted travel planning platform that turns swipe-based preferences into personalized itineraries with automatic calendar synchronization.

Itineraries in <30 seconds60% less manual schedulingGoogle Maps + Calendar APIs

Live Site →GitHub

Overview

Tratlus is a travel planning platform that converts vague preferences into concrete, schedulable itineraries. Users swipe through destination options, and the system generates personalized travel plans that automatically sync with their calendars.

Problem

Planning travel is tedious. Users spend hours researching destinations, cross-referencing calendars, and manually creating schedules. Most planning tools require significant upfront input before providing any value. The result: abandoned plans and decision fatigue.

Constraints

Preference capture needed to be fast and low-friction (mobile-first)
Generated plans had to account for real calendar constraints
System needed to handle payment flow for booking fees
API rate limits on Google Maps and Calendar

Solution

I built a swipe-based preference capture system that uses a swipe-based interface to minimize friction during preference capture. Once preferences are collected, LLM-driven prompt engineering generates structured itineraries. The system then integrates with Google Calendar to automatically schedule activities and prevent conflicts, reducing manual scheduling work by 60%.

Architecture & Stack

FrontendReact, TypeScript, mobile-first responsive UI

BackendFirebase Cloud Functions

DatabaseFirebase Firestore

External APIsGoogle Maps, Google Calendar, Stripe

Key Engineering Decisions

Swipe-based preference capture to reduce cognitive load
Structured LLM prompts producing JSON-formatted itineraries
Google Calendar integration for conflict-aware scheduling
Stripe payment flow with webhook-based confirmation

Challenges & Tradeoffs

Coordinating multiple external APIs (Maps, Calendar, Stripe) while keeping the UX snappy was the core challenge. I implemented optimistic UI updates and background syncing—the user sees immediate feedback while the system reconciles with external services asynchronously. The tradeoff: eventual consistency instead of strong consistency, but significantly better perceived performance.

Results

<30s

Itinerary generation

60%

Less manual scheduling

Zero

Calendar conflicts

What This Demonstrates

API orchestration across multiple external services
Payment integration with proper webhook handling
UX design that reduces friction and increases task completion
LLM prompt engineering for structured, parseable outputs